Accepted Manuscript

A hybrid modeling approach for predicting the educational use of mobile cloud
computing services in higher education

Ibrahim Arpaci

PII: S0747-5632(18)30445-X

DOI: 10.1016/j.chb.2018.09.005

Reference: CHB 5696

To appear in: Computers in Human Behavior

Received Date: 20 January 2018

Accepted Date: 10 September 2018

Please cite this article as: Ibrahim Arpaci, A hybrid modeling approach for predicting the
educational use of mobile cloud computing services in higher education, Computers in Human
Behavior (2018), doi: 10.1016/j.chb.2018.09.005

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form.
Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

A hybrid modeling approach for predicting the educational use of mobile cloud computing

services in higher education

Ibrahim Arpaci, Ph.D.

Tokat Gaziosmanpasa University, Department of Computer Education and Instructional

Technology, Tokat, Turkey, ibrahim.arpaci@gop.edu.tr

ABSTRACT

The decision to integrate mobile cloud computing (MCC) in education without determining optimal

use scenarios is a universal problem as the adoption of such services becomes widespread.

Accordingly, this study developed and validated a predictive model that explains the role of

students’ information management (i.e. retrieve, store, share, and apply) practices in predicting

their attitudes toward using the MCC services for educational purposes. This study validated the

model by the complementary use of machine learning algorithms alongside a classical SEM-based

approach based on data collected from 308 undergraduate students. The SEM results indicated that

the students’ information management (i.e. retrieve, store, share, and apply) practices were

significantly associated with their attitudes, which were significantly associated with the behavioral

intentions. The structural model explained a significant portion of the variance in the behavioral

intentions. Likewise, the classifier model suggested that the students’ information management

practices and attitudes predicted the behavioral intentions. Further, the applied algorithms predicted

the behavioral intentions with an accuracy of more than 72% in most cases. Thereby, the study

extended an original theory (TRA) into the MCC area by using a multi-analytical approach. The

findings implied that employing the MCC services for personal information management should

be supported and encouraged in the higher education by designing authentic learning environments
 ACCEPTED MANUSCRIPT

and scaffolding the students in using such services.

Keywords: information management; cloud computing; artificial intelligence; machine learning

 ACCEPTED MANUSCRIPT

A hybrid modeling approach for predicting the educational use of mobile cloud computing

services in higher education

ABSTRACT

The decision to integrate mobile cloud computing (MCC) in education without determining optimal

use scenarios is a universal problem as the adoption of such services becomes widespread.

Accordingly, this study developed and validated a predictive model that explains the role of

students’ information management (i.e. retrieve, store, share, and apply) practices in predicting

their attitudes toward using the MCC services for educational purposes. This study validated the

model by the complementary use of machine learning algorithms alongside a classical SEM-based

approach based on data collected from 308 undergraduate students. The SEM results indicated that

the students’ information management (i.e. retrieve, store, share, and apply) practices were

significantly associated with their attitudes, which were significantly associated with the behavioral

intentions. The structural model explained a significant portion of the variance in the behavioral

intentions. Likewise, the classifier model suggested that the students’ information management

practices and attitudes predicted the behavioral intentions. Further, the applied algorithms predicted

the behavioral intentions with an accuracy of more than 72% in most cases. Thereby, the study

extended an original theory (TRA) into the MCC area by using a multi-analytical approach. The

findings implied that employing the MCC services for personal information management should

be supported and encouraged in the higher education by designing authentic learning environments

and scaffolding the students in using such services.

Keywords: information management; cloud computing; artificial intelligence; machine learning

 ACCEPTED MANUSCRIPT
2

1. Introduction

Mobile cloud computing (MCC) as a new distributed computing paradigm can be defined as an

infrastructure, application or process, where the data storage and processing migrated from smart

mobile technologies to the distributed cloud servers (Dinh, Lee, Niyato, & Wang, 2013). The MCC

has gained a substantial attention of both the organizations and individuals as a promising solution

for the ubiquitous environments, in which data storage and processing occur over a “cloud” via the

Internet (Park & Kim, 2014). Investing in the MCC, organizations improved their capacity and

capabilities without the cost of installing new infrastructure or software (Subashini & Kavitha,

2011). Further, the MCC reduced the initial installation and maintenance costs, and thereby,

promoted the efficiency and green technology (Aepona, 2018).

There are various cloud deployment-models such as private, public, hybrid, and community cloud

(Shon, Cho, Han, & Choi, 2014). The study focused on the public cloud that allocates resources on

a per-user basis through applications such as Dropbox, SkyDrive, and Google Drive. However, the

MCC service models for individuals include “platform as a service” (PaaS) for automatic content

synchronizing and “infrastructure as a service” (IaaS) for information storage and management.

On the other hand, these services provide enterprises with the IaaS for an enterprise infrastructure

building and “software as a service” (SaaS) for an enterprise solution provider (Arpaci, 2016).

The MCC provides users at the organizational and individual level with several advantages and

given these advantages the MCC services are considered to be the growth engine of the industry

4.0 (Park & Kim, 2014). For example, the MCC increases data storage capacity and allows an

efficient data synchronization and information management in a ubiquitous environment (Park &

Kim, 2014). Besides, energy efficiency is a critical issue since battery power of the mobile

technologies is limited. Migration of the complex processing from mobile technologies to the cloud
 ACCEPTED MANUSCRIPT
3

servers extends the battery life (Cuervo et al., 2010). Furthermore, service providers have certain

security mechanisms and backup systems to protect user data. Therefore, saving documents and

files on the distributed cloud servers may enhance the reliability by reducing virus-like threats and

data loss (Arpaci, 2016).

Smart mobile technologies change the way of information is managed (Ogiela, 2017). For example,

using mobile Internet on smartphones provides students with ubiquitous access to information and

learning materials. Further, social networking and communication applications allow them to share

information with peers. More importantly, the integrated technologies enable them to use

information stored in the MCC services in problem solving and decision-making. Considering the

extensive use of mobile devices such as smartphones, tablets, and smartwatches by undergraduate

students, it is important to find creative ways for an effective integration of these technologies into

the higher education. Such an integration may enable students to learn by using the technologies,

which they are familiar and confident with, and highly motivated to use. It is important to note that

providing an effective management of personal information through information retrieval, access,

storage, sharing, and application is one of the main advantages of the MCC services. However, the

decision to integrate the MCC in educational settings without determining optimal use scenarios is

a universal problem as the adoption of such services becomes widespread. Thus, this study

investigated adoption of the MCC among undergraduate students by focusing on the use of these

services for personal information management.

2. Literature Review

The MCC is a recent research area and there are numerous studies on the adoption of the MCC at

the both organizational and individual level. For example, Ul Amin, Inayat, Shahzad, Saleem, and

Aijun (2017) applied the “Unified Theory of Acceptance and Use of Technology” (UTAUT) to
 ACCEPTED MANUSCRIPT
4

identify key predictors of the MCC adoption by healthcare professionals. Their results suggested

the UTAUT constructs (i.e. performance-expectancy, social influence, effort-expectancy) were

significant in determining the behavioral intentions. In another study, Lian, Yen, and Wang (2014)

investigated the factors affecting the cloud computing adoption in Taiwan’s hospital industry based

on the HOT-fit (Human-Organization-Technology fit) model and TOE (Technology-Organization-

Environment) framework. The results suggested that technical competence, data security, cost,

complexity, and top management support were critical factors in the adoption. Chen, Chen, and

Lee (2018) investigated key factors affecting the organizations’ adoption of the cloud services.

Their results suggested that top-management-support has a vital role in the adoption at the

organizational level.

Lee (2016) identified the key factors explaining the adoption of cloud services at the individual

level based on the “Diffusion of Innovations” (DOI) theory. The results suggested the relative

advantage, observability, and self-efficacy were significant antecedents of the adoption. Park and

Kim (2014) focused on key factors predicting the individuals’ acceptance of the MCC services.

Results indicated that connectedness, perceived mobility, service quality, security, and satisfaction

were significant predictors of the user acceptance. Wang and Huang (2016) reported that social

influence was a key predictor of the students’ intention to use cloud services. Further, they reported

that scaffolding in problem-solving and training were useful in familiarizing the students with such

services.

Arpaci (2016) proposed a theoretical structural model based on the “Technology Acceptance

Model” (TAM) and suggested that subjective norm, trust, and perceived-usefulness play significant

roles in predicting the students’ attitudes towards using the MCC services. In another study,

Sharma, Al-Badi, Govindaluri, and Al-Kharusi (2016) developed a hybrid model based on the
 ACCEPTED MANUSCRIPT
5

TAM to predict factors affecting the adoption of the cloud computing by IT professionals. They

employed linear regression along with neural network analysis and found that job opportunity was

the most significant predictor of the adoption. Shiau and Chau (2016) developed a theoretical

structural model based on the TAM and motivational model to predict behavioral intentions to join

a cloud computing class. The results suggested subjective norms and attitudes were positively

associated with the behavioral intention. Kim and Kim (2018) proposed a research model in order

to identify the antecedents of the MCC adoption. The results suggested that trust, convenience, and

perceived uncertainty were significant factors in the individuals’ adoption decision.

3. Theoretical Background

Most of the studies reviewed applied or extended a theory such as TAM (Davis, 1989) or UTAUT

(Venkatesh, Morris, Davis, & Davis, 2003) by conducting an explanatory statistical analysis.

However, such an approach has two main limitations. First, each theory was developed and tested

for a specific domain, and therefore, may not work well in other domains or areas. Thus, this study

extended the original “Theory of Reasoned Action” (TRA) with additional constructs to better

explain the research context. Second, previous studies pointed out statistical limitations of the

technology adoption studies and emphasized the importance of employing a multi-analytical

approach by combining predictive analytics and causal explanatory statistical analysis to validate

a predictive model (Sharma et al., 2016; Sharma, Joshi, & Sharma, 2016; Tan, Ooi, Leong, & Lin,

2014). The study therefore tested the research model by using both the classical SEM approach and

artificial intelligence techniques (i.e. machine learning). The study employed the SEM approach to

understand causal relationships and the (complementary) study applied the machine learning

algorithms to predict the behavioral intentions based on the proposed constructs. Further, this study

approaches the MCC adoption from an information management perspective. The fact that
 ACCEPTED MANUSCRIPT
6

providing an effective information management was one of the main advantages of the MCC

services supports the idea that individuals’ information management practices may have a critical

role in their adoption decision. Accordingly, this study investigated the role of students’

information management practices on their attitudes toward using the MCC services in educational

settings.

The TRA postulated that beliefs affect attitudes, which positively influence behavioral intentions,

whereas the intentions ultimately influence the actual behavior (Fishbein & Ajzen, 1975). Benbasat

and Barki (2007) suggested using the original theory (TRA) to provide a strong theoretical

grounding for a novel model by incorporating different antecedents relevant to the nature of ICT

integration and use in diverse settings. Likewise, Nistor (2014) argued that “educational technology

acceptance” studies should consider characteristics of the educational context. Considering the

shortcomings of the technology acceptance models and analysis methods suggested by (Benbasat

& Barki, 2007), this study identified the key antecedents of the attitudes towards using the MCC

services and employed an innovative approach comprising both machine learning algorithms and

the SEM, and thereby, strengthened the study design. Figure 1 illustrates the research model that

suggests the behavioral intentions (continued use intentions) are predicted by the attitudes, which

are predicted by the students’ information management practices such as information retrieval,

storage, sharing, and application.

Rogers’s (2003, p.19) DOI theory suggested that the stages by which an individual adopts a new

technology or innovation, whereby diffusion is accomplished include “awareness of the need for

an innovation, decision to adopt (or reject) the innovation, initial use of the innovation to test it,

and continued use of the innovation.” Accordingly, a training program introduced the participants

to personal information management, which refers to the practice of activities an individual

 ACCEPTED MANUSCRIPT
7

performs to create or acquire, store, retrieve, share, and apply the information required to complete

the tasks. The participants employed the mobile web services (i.e. search, maps), mobile cloud

applications (i.e. Google Drive, iCloud, Dropbox) and social media applications to perform the

training tasks.

3.1. Hypotheses

3.1.1. Behavioral Intentions and Attitudes

“Behavioral intention” was defined as “the degree of an individual’s belief that he or she will

continue to use a system” (Arpaci, 2017, p. 384). However, attitudes towards using a technology

was defined as “an individual’s overall affective reaction to use a particular system” (Arpaci, 2017,

p. 384). This study therefore hypothesized that the attitude toward using the MCC services would

be positively associated with the behavioral intention to use such services for educational purposes.

Further, the attitude would be a predictor of the behavioral intention. Therefore;

H1a. Attitude towards using the MCC services would be positively associated with the behavioral

intention.

H1b. Attitude towards using the MCC services would predict the behavioral intention.

3.1.2. Information Management Practices

Information is processed, categorized, organized or structured data (Nonaka, 1994; Davenport &

Prusak, 2000). However, knowledge is a combination of skills, information, experience, and

insights (Nonaka & Takeuchi, 1995). Knowledge includes organizational routines, values, and

practices; therefore, this study preferred to use the term “information management” instead of

“knowledge management.” This study defined information management as practices that help

retrieve and store information, share it with others, and apply it to solve problems and make better
 ACCEPTED MANUSCRIPT
8

and timely decisions (Alavi & Leidner, 2001; Mitchell, 2003). Turban, Sharda, and Delen (2011)

suggested the use of information technologies to enhance information management practices of

retrieval and access, transfer, storage, and application. In the same vein, the present study suggested

the integration of smart mobile technologies into the higher education to support personal

information management. The more information managed by using the MCC services the more

favorable the attitude towards using such services for educational purposes. Further, the students’

information management practices would be a predictor of the behavioral intention to use the MCC

services for educational purposes. Accordingly,

H2a. An increase in the volume of information that’s retrieved via the use of the MCC services is

positively associated with the attitude towards the use of such services.

H2b. Information retrieval would predict the behavioral intention to use the MCC services for

educational purposes.

H3a. An increase in the volume of information that’s stored via the use of the MCC services is

positively associated with the attitude towards the use of such services.

H3b. Information storage would predict the behavioral intention to use the MCC services for

educational purposes.

H4a. An increase in the volume of information that’s shared via the use of the MCC services is

positively associated with the attitude towards the use of such services.

H4b. Information sharing would predict the behavioral intention to use the MCC services for

educational purposes.
 ACCEPTED MANUSCRIPT
9

H5a. An increase in the volume of information that’s applied via the use of the MCC services is

positively associated with the attitude towards the use of such services.

H5b. Information application would predict the behavioral intention to use the MCC services for

educational purposes.

4. Method

4.1. Sample

A sample of 308 university students were recruited from a public university in Turkey. The study

included 186 females (60.4%) and 122 (39.6%) males with a mean age of 21.88 years (SD=2.37).

Majority of the students (42.5%) were seniors; 12.3% were freshmen, 27.6% were sophomore, and

17.5% were juniors.

4.2. Procedure

IRB of the affiliated university approved the research and the procedures complied with the ethical

standards of the institutional board guidelines. All participants were informed about the purpose of

the research after an informed consent was obtained. The study was conducted in regularly

scheduled IT classes (4 hours per week) during 14 weeks. The participants performed practical

implementations of the theoretical knowledge during the interactive tasks such as: 1) Search for

research articles in electronic databases (i.e. Scopus, WOS, and Google Scholar), 2) Save an

annotated bibliography of the articles to the MCC services (i.e. OneDrive, Dropbox, and Google

Drive), 3) Form a communities of practice and use the MCC services along with social networks

to share files, communicate with others, and build a common cognition to complete the tasks.

Finally, the survey instrument with explicit assurance that anonymity is guaranteed was

administered during the last week of the regular classes.

 ACCEPTED MANUSCRIPT
10

4.3. Instruments

The measurement instrument for the variables “behavioral intentions” and “attitudes” was a Likert

scale adapted from Ajzen’s theory (1991). Moreover, the measurement instrument for the

information management practices of retrieval, storage, sharing, and application was adapted from

a previously validated and reliable instrument (Arpaci, 2017). The final instrument has 24 items

with a five-point Likert-type scale ranging from “strongly disagree” to “strongly agree.”

5. Results

5.1. Internal Reliability and Validity

IBM SPSS (ver. 25) and IBM AMOS (ver. 23) were used to screen and analyze the data. Construct

validity along with reliability of the scales were assessed by conducting an EFA (See Table 1).

Results revealed that each item had a communality value and a factor loading greater than the

critical threshold (.40) suggested by Field (2005). Cronbach’s alpha reliability analysis results

indicated a good homogeneity and reliability among the items. The AVE values were greater than

.50, indicating adequate convergent validity for all constructs (Hair, Tatham, Anderson, & Black,

2006).

[Table 1 will be about here]

5.2. The Structural Model

Kaiser’s (1970) “KMO measure of sampling adequacy” and Bartlett’s (1951) “test of sphericity”

suggested that conducting a CFA on the data was appropriate (See Table 2). KMO values were all

above .65 and Bartlett’s test values were all significant (p< .001). This suggested that measures for

each construct were interdependent (Leech, Barrett, & Morgan, 2005).

[Table 2 will be about here]

 ACCEPTED MANUSCRIPT
11

The CFA was conducted using maximum-likelihood estimates in an attempt to test the research

model. Results suggested the structural model demonstrates a “good fit” to data: [χ2/df=1.67,

GFI=.91, NFI=.92, AGFI=.88, CFI=.96, TLI=.96, IFI=.96, RMSEA=.05]. A GFI, NFI, TLI, CFI,

and IFI equal or above .90 and a RMSEA less than .08 indicate an acceptable model fit (Hair et al.,

2006). Further, Kline (2005) suggested a χ2/df less than 2.0 indicates a good fit.

5.3. Hypothesis Testing

A SEM approach was used to explore causal relationships among the variables. The standardized-

path-coefficients in the structural model were all statistically significant (See Figure 1). Results

indicated that information retrieval, information storage, information sharing, and information

application were significantly associated with the attitudes toward using the MCC services. All

these factors explained .33 of the variance in the attitudes. Whereas, the information retrieval had

the largest impact on the attitudes with a standardized coefficient of .45. On the other hand, the

information storage had a relatively small impact on the attitudes with a standardized coefficient

of .16.

[Figure 1 will be about here]

The results shown in Table 3 suggested the attitudes were significantly associated with the

behavioral intentions with a standardized coefficient of .77. These factors accounted for a

significant variance in explaining the behavioral intentions (R2 =.59).

[Table 3 will be about here]

5.4. Testing the Predictive Model using the Machine Learning Algorithms

Classification is one of the machine learning techniques that aims to build a classifier model in

order to predict a behavior through classifying the data into a number of pre-defined classes based
 ACCEPTED MANUSCRIPT
12

on a certain criterion (Ngai, Xiu, & Chau, 2009). Common techniques employed for the

classification are Bayesian networks, decision trees, if-then-else rules, and neural networks.

Machine learning, an artificial/computational intelligence technique, uses a set of attributes and

searches for correlations between the attributes and performance of the learning algorithms

(Kotsiantis, Zaharakis, & Pintelas, 2007). This study employed machine learning classification

algorithms to predict the behavioral intentions based on attributes of the students’ information

management practices and their attitudes. Behavioral intentions and attitudes have nominal values

and classified as low, medium or high by using the Mean ± SD combination. On the other hand,

the attributes of information retrieval, storage, sharing, and application have numeric values.

The predictive model was tested by using Weka (ver. 3.8.1) based on the most common

classification algorithms, which classify data into a number of pre-defined classes with different

learning schemas. The results shown in Table 4 provided performance of the best performance

classifier based on two test modes; 10-fold cross-validation and percentage split (66.0%).

[Table 4 will be about here]

The study employed a linear logistic regression classifier (Logistic), a meta AdaBoostM1, a

Bayesian classifier (Naive bayes), a lazy LWL, and two rule learners (OneR, JRip). The results

showed that the OneR rule learner performs slightly better than the other classifiers. OneR classifier

predicted the behavioral intentions with an accuracy of 73.70% and 71.43% for 10-fold cross-

validation and percentage split, respectively. Further, this algorithm had a better performance in

terms of true positive (TP) rate (.74) and F-measure (.73) for the 10-fold test mode compared to

the other classifiers (i.e. Naive bayes, Logistic, JRip, AdaBoostM1, LWL). The results shown in

Table 5 suggested that attributes of the attitudes, information retrieval, information storage,
 ACCEPTED MANUSCRIPT
13

information sharing, and information application predicted the behavioral intentions with an

accuracy of 74%, and thereby, H1b, H2b, H3b, H4b, and H5b were supported.

[Table 5 will be about here]

6. Discussion and Conclusion

6.1. Research Implications

Using the MCC services provides students with ubiquitous access to instructional materials and

information anytime and anywhere. In addition, these services provide students with a storage

capacity to store their reference materials. Furthermore, the MCC services enable the students to

share ideas, documents, and files with peers and classmates. Eventually, the students can apply and

transform that knowledge and experience in problem-solving and decision-making. However, the

decision to integrate these services in educational settings requires determining an optimal use

scenario. This study therefore focused on the use of the MCC services for personal information

management. Accordingly, the study employed a hybrid modeling approach for investigating the

role of students’ information management (i.e. retrieve, store, share, and apply) practices in their

attitudes (toward use) and the intentions to (continue to) use the MCC services.

The actuality of the ideas and employing an integrated multi-analytical approach were the novel

contributions of the study. The study developed a research model by extending the TRA with

additional constructs and tested the model by integrating the SEM and machine learning approach.

This study identified the causal relationships between the dependent and independent variables by

employing the SEM approach. The SEM results indicated that the students’ information

management practices (i.e. retrieve, store, share, and apply) and attitudes were significantly

associated with the behavioral intentions to use the MCC services for educational purposes.
 ACCEPTED MANUSCRIPT
14

In a complementary study, machine learning algorithms were employed to explore whether the

students’ information management practices and attitudes would predict the behavioral intentions.

The study tested a number of classification algorithms with different methodologies, including

bayesian networks, decision trees, if-then-else rules, association rules, and neural networks. The

results indicated that the OneR rule learner performs slightly better than other classifiers. It is

worthy to mention that OneR (or one class) algorithm works well if one of the attributes can predict

the outcome (dependent) variable better than the other attributes in the training data (Holte, 1993).

The classifier model suggested that the attitude predicts the behavioral intention better than the

other attributes (CCI=.74). In line with these results, the structural model suggested that the attitude

was a significant factor in predicting the behavioral intention (β=.77). Further, the SEM results

suggested the students’ information management practices and attitudes were significantly

associated with the behavioral intentions and accounted for 59% of the variance in the behavioral

intentions.

It is important to note that the relation between the attitudes and the behavioral intentions explains

a lot of variance in the model, which is desirable, but not informative or useful. Conceptually, the

attitudes (toward use) and the intentions to (continue to) use are a (conceptual) component of the

later. However, this study overcomes this design flaw by capitalizing on its two main strengths: 1)

the complementary use of parametric and nonparametric approaches, and 2) the extension of the

original theory (TRA) into the MCC area.

The SEM model included information management practices as exogenous (independent)

variables. In the model, both the attitudes and behavioral intentions were endogenous (dependent)

variables. The SEM results identified the causal relationships among the endogenous variables and

between the endogenous and exogenous variables. On the other hand, the classifier model included
 ACCEPTED MANUSCRIPT
15

the attitudes along with information management practices as predictors of the behavioral

intentions. Although, presentation of the two analyses (SEM vs. machine learning) seems two

separate results, considering the operationalization of the same constructs in both model, indeed,

this study presents a “hybrid modelling” approach and not a mere dual-perspective approach.

6.2. Practical Implications

An effective management of information has a potential to improve the students’ academic

performance and efficiency of the learning tasks. This implies that employing the MCC services

for personal information management should be supported and encouraged in the higher education

by designing authentic learning environments and scaffolding students in using such services.

Universities should develop their own policies (i.e. bring-your-own-device policy) and cloud-based

applications to maximize benefits of the such services. More importantly, instructors should

employ blended learning strategies that require information management practices to promote

educational use of the MCC services.

6.3. Limitations and Future Directions

The SEM does not provide a complete solution to all statistical problems of the prediction. On the

other hand, it provides a powerful means of testing a causal model based on a well-specified

conceptual theory. Accordingly, the complementary use of the SEM and the machine learning

algorithms to develop a prediction model provided unique information on the performance of the

SEM-based prediction compared to the other prediction models. Besides, using two approaches,

one focused on contrasting/comparing and another on completing, help make robust predictions.

However, the study has a number of limitations. This study did not account for the potential

moderating role of cultural orientations in the educational technology acceptance, as reported by

(Arpaci, 2015; Arpaci, Cetin Yardimci, & Turetken, 2015). Therefore, one should be careful to
 ACCEPTED MANUSCRIPT
16

generalize the findings to the samples with a different socio-cultural background. Concerns

reported by (Stergiou & Psannis, 2017) may also affect the degree of adoption. Future studies may

therefore focus on the concerns of confidentiality, privacy, latency, performance instability, data

security, and lack of service-level-agreements. The fact that an effective technology integration

into the higher education requires a techno-pedagogical approach and needs to consider the aspects

of pedagogy, curriculum, change management, competencies, and organizational readiness,

implies that future studies should take into account all these aspects. Finally, a longitudinal study

is recommended to overcome limitations of the cross-sectional design.

 ACCEPTED MANUSCRIPT
17

References

Aepona, 2018. Mobile Cloud Solutions. Retrieved from http://www.aepona.com/solutions/mobile-

cloud/ (accessed July 16, 2018).

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision

Processes, 50(2), 179-211.

Alavi, M., & Leidner, D. E. (2001). Review: Knowledge management and knowledge management

systems: Conceptual foundations and research issues. MIS Quarterly, 25(1), 107-136.

Arpaci, I. (2015). A comparative study of the effects of cultural differences on the adoption of

mobile learning. British Journal of Educational Technology, 46(4), 699-712. Doi:

10.1111/bjet.12160

Arpaci, I. (2016). Understanding and predicting students’ intention to use mobile cloud storage

services. Computers in Human Behavior, 58, 150-157. Doi: 10.1016/j.chb.2015.12.067

Arpaci, I. (2017). Antecedents and consequences of cloud computing adoption in education to

achieve knowledge management. Computers in Human Behavior, 70(5), 382-390. Doi:

10.1016/j.chb.2017.01.024

Arpaci, I., Cetin Yardimci, Y., & Turetken, O. (2015). A cross-cultural analysis of smartphone

adoption by Canadian and Turkish organizations. Journal of Global Information Technology

Management, 18(3), 214-238. Doi:10.1080/1097198X.2015.1080052.

Bartlett, M. S. (1951). The effect of standardization on a χ2 approximation in factor analysis.

Biometrika, 38(3/4), 337-344.

 ACCEPTED MANUSCRIPT
18

Benbasat, I., & Barki, H. (2007). Quo vadis TAM? Journal of the Association for Information

Systems 8(4), 211-218.

Chen, S. L., Chen, J. H., & Lee, Y. H. (2018). A comparison of competing models for

understanding industrial organization’s acceptance of cloud services. Sustainability, 10(3), 673.

Cuervo, E., Balasubramanian, A., Cho, D. K., Wolman, A., Saroiu, S., Chandra, R., & Bahl, P.

(2010). MAUI: Making smartphones last longer with code offload. In Proceedings of the 8th

international conference on Mobile systems, applications, and services (pp. 49-62). ACM.

Davenport, T. H., & Prusak, L. (2000). Working knowledge: How organizations manage what they

know. Boston, MA: Harvard Business School Press.

Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of

information technology. MIS Quarterly, 13(3), 319-340.

Dinh, H. T., Lee, C., Niyato, D., & Wang, P. (2013). A survey of mobile cloud computing:

architecture, applications, and approaches. Wireless Communications and Mobile Computing,

13(18), 1587-1611.

Field, A. (2005). Discovering statistics using SPSS. Thousand Oaks, CA: SAGE.

Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to theory

and research. Addison-Wesley, Reading, MA.

Hair, J. F., Tatham, R. L., Anderson, R. E., & Black, W. (2006). Multivariate data analysis (6th

Ed.). Upper Saddle River, NJ: Pearson Prentice Hall.

 ACCEPTED MANUSCRIPT
19

Holte, R. C. (1993). Very simple classification rules perform well on most commonly used datasets.

Machine Learning, 11(1), 63-90.

Kaiser, H. F. (1970). A second generation little jiffy. Psychometrika, 35(4), 401-415.

Kim, S. H., & Kim, J. K. (2018). Determinants of the adoption of mobile cloud computing services:

a principal-agent perspective. Information Development, 34(1), 44-63.

Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). New York:

The Guilford Press.

Kotsiantis, S. B., Zaharakis, I., & Pintelas, P. (2007). Supervised machine learning: A review of

classification techniques. Emerging Artificial Intelligence Applications in Computer Engineering,

160, 3-24.

Lee, Y. C. (2016). Why do people adopt cloud services? Gender differences. Social Science

Information, 55(1), 78-93.

Leech, N. L., Barrett, K. C., & Morgan, G. A. (2005). SPSS for intermediate statistics: Use and

interpretation (2nd Ed.). Hillsdale, NJ: Lawrence Erlbaum.

Lian, J. W., Yen, D. C., & Wang, Y. T. (2014). An exploratory study to understand the critical

factors affecting the decision to adopt cloud computing in Taiwan hospital. International Journal

of Information Management, 34(1), 28-36.

Mitchell, H. J. (2003). Technology and knowledge management: Is technology just an enabler or

does it also add value? In Knowledge Management: Current Issues and Challenges (pp. 66-78).

Hershey: IRM Press.

 ACCEPTED MANUSCRIPT
20

Ngai, E. W., Xiu, L., & Chau, D. C. (2009). Application of data mining techniques in customer

relationship management: A literature review and classification. Expert Systems with Applications,

36(2), 2592-2602.

Nistor, N. (2014). When technology acceptance models won’t work: Non-significant intention-

behavior effects. Computers in Human Behavior, 34, 299-300. Doi: 10.1016/j.chb.2014.02.052

Nonaka, I. (1994). A dynamic theory of organizational knowledge creation. Organization Science,

5(1), 14-37.

Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating company: How Japanese companies

create the dynamics of innovation. New York: Oxford University Press.

Ogiela, M. R. (2017). Impact of ubiquitous computing technologies for modern information

management. International Journal of Information Management, 1(37), 1517-1518.

Park, E., & Kim, K. J. (2014). An integrated adoption model of mobile cloud services: Exploration

of key determinants and extension of technology acceptance model. Telematics and Informatics,

31(3), 376-385.

Rogers, E. M. (2003). Diffusion of innovations (5th ed.). New York, NY: Free Press.

Sharma, S. K., Al-Badi, A. H., Govindaluri, S. M., & Al-Kharusi, M. H. (2016). Predicting

motivators of cloud computing adoption: A developing country perspective. Computers in Human

Behavior, 62, 61-69.

Sharma, S. K., Joshi, A., & Sharma, H. (2016). A multi-analytical approach to predict the Facebook

usage in higher education. Computers in Human Behavior, 55, 340-353.

 ACCEPTED MANUSCRIPT
21

Shiau, W. L., & Chau, P. Y. (2016). Understanding behavioral intention to use a cloud computing

classroom: A multiple model comparison approach. Information & Management, 53(3), 355-365.

Shon, T., Cho, J., Han, K., & Choi, H. (2014). Toward advanced mobile cloud computing for the

internet of things: current issues and future direction. Mobile Networks and Applications, 19(3),

404-413.

Stergiou, C., & Psannis, K. E. (2017). Recent advances delivered by mobile cloud computing and

internet of things for big data applications: A survey. International Journal of Network

Management, 27(3), e1930.

Subashini, S., & Kavitha, V. (2011). A survey on security issues in service delivery models of

cloud computing. Journal of Network and Computer Applications, 34(1), 1-11.

Tan, G. W. H., Ooi, K. B., Leong, L. Y., & Lin, B. (2014). Predicting the drivers of behavioral

intention to use mobile learning: A hybrid SEM-Neural Networks approach. Computers in Human

Behavior, 36, 198-213.

Turban, E., Sharda, R., & Delen, D. (2011). Decision support and business intelligence systems.

Boston: Prentice Hall.

Ul Amin, R., Inayat, I., Shahzad, B., Saleem, K., & Aijun, L. (2017). An empirical study on

acceptance of secure healthcare service in Malaysia, Pakistan, and Saudi Arabia: a mobile cloud

computing perspective. Annals of Telecommunications, 72(5-6), 253-264.

Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information

technology: Toward a unified view. MIS Quarterly, 27(3), 425-478

 ACCEPTED MANUSCRIPT
22

Wang, C. S., & Huang, Y. M. (2016). Acceptance of cloud services in face-to-face computer-

supported collaborative learning: A comparison between single-user mode and multi-user mode.

Innovations in Education and Teaching International, 53(6), 637-648.

 23

Table 1. Instrument reliability and validity

Construct Item α Item-total Factor Comm Total CR AVE

correlation load unality variance

explained

Behavioral CUI1. “I intend to continue to use mobile .91 .76 .85 .72 74.34 .91 .68

intention cloud services for educational purposes in the

future.”

CUI2. “I plan to continue to use mobile cloud .78 .86 .75

services for personal information

management in the future.”

CUI3. “I predict that I would frequently use .77 .86 .74

mobile cloud services in the future.”

CUI4. “I predict that I would frequently use .79 .87 .76

mobile cloud services for educational

purposes.”
 24

CUI5. “I predict that I would frequently use .79 .87 .76

mobile cloud services for personal

information management.”

Attitude A1. “Using mobile cloud services for .89 .73 .82 .67 65.30 .89 .59

educational purposes is a good idea.”

A2. “Using mobile cloud services to create a .78 .86 .74

retrievable archive of personal information is

pleasant.”

A3. “Using social media applications on my .67 .78 .61

smartphone to exchange information is fun.”

A4. “Using mobile cloud services for .75 .83 .69

personal information management is a wise

idea.”
 25

A5. “Using mobile cloud services for .68 .78 .61

educational purposes makes learning more

interesting.”

A6. “I like the idea of using mobile cloud .67 .77 .60

services for personal information

management.”

Informatio IR1. “Using mobile Internet on my .91 .75 .86 .74 78.43 .91 .51

n retrieval smartphone enables to retrieve learning

materials and information anytime and

anywhere.”

IR2. “I access course materials and .85 .92 .84

information using mobile cloud services on

my smartphone.”
 26

IR3. “Using mobile cloud services on my .84 .92 .84

smartphone enables quick access to learning

materials and information resources.”

IR4. “Using mobile cloud services on my .74 .85 .72

smartphone enables ubiquitous access to e-

databases and e-journals.”

Informatio IS1. “Using mobile cloud services on my .91 .81 .91 .83 84.06 .91 .52

n storage smartphone enables the retrievable storage of

electronic information and documents.”

IS2. “I store course materials and documents .83 .93 .86

using mobile cloud services on my

smartphone.”

IS3. “Using mobile cloud services enables .80 .91 .83

me to store learning materials and

information with ubiquitous access.”

 27

Informatio ISh1. “Using social media applications on my .71 .51 .80 .64 64.61 .71 .51

n sharing smartphone enables me to exchange

information and documents with classmates.”

ISh2. “I share course materials and .62 .86 .74

information with classmates using mobile

cloud services on my smartphone.”

ISh3. “I share learning materials and .48 .75 .57

documents with classmates using mobile

cloud services on my smartphone.”

Informatio IA1: “I apply knowledge and experience .83 .64 .83 .69 74.72 .83 .53

n gained by using mobile cloud services to

application complete learning tasks.”

IA2. “I use knowledge obtained by using .73 .89 .79

mobile cloud services in decision making

processes.”
 28

IA3. “I employ knowledge and intelligence .71 .88 .67

gained from using mobile cloud services in

problem‐solving activities.”
 ACCEPTED MANUSCRIPT
29

Table 2. KMO and Bartlett’s test results

KMO Chi-Square Sig.

Behavioral intention .88 1022.27 .001

Attitude .88 1001.80 .001

Information retrieval .83 839.40 .001

Information storage .75 594.58 .001

Information sharing .65 198.79 .001

Information application .71 357.46 .001

 ACCEPTED MANUSCRIPT
30

Table 3. The SEM results

Hypothesis Path Estimate Std. Std. Critical p-value Remarks

Estimate Error Ratio

H1a A→BI .983 .776 .098 10.04 .001 Supported

H2a IR→A .412 .446 .059 6.976 .001 Supported

H3a ISt→A .084 .156 .030 2.802 .005 Supported

H4a ISh→A .125 .180 .043 2.918 .004 Supported

H5a IA→A .207 .279 .045 4.605 .012 Supported

Note: A= Attitude; BI= Behavioral Intention; IR= Information Retrieval; ISt= Information Storage;

ISh= Information Sharing; IA= Information Application

 ACCEPTED MANUSCRIPT
31

Table 4. Detailed results for the OneR rule learner

TP FP Precision Recall F- MCC ROC PRC Class

Rate Rate Measure Area Area

10-fold .648 .089 .687 .648 .667 .571 .780 .526 Low

cross- .845 .430 .770 .845 .806 .434 .708 .748 Medium

validation .395 .042 .607 .395 .479 .427 .677 .324 High

Avg. .737 .297 .728 .737 .728 .465 .720 .638

Percentage .650 .118 .565 .650 .605 .505 .766 .434 Low

split .800 .429 .789 .800 .794 .374 .686 .764 Medium

(66%) .400 .056 .545 .400 .462 .394 .672 .304 High

Avg. .714 .316 .711 .714 .711 .402 .699 .636

 ACCEPTED MANUSCRIPT
32

Table 5. Performance results for the applied classifiers

Algorithm CCI1 (%) TP2-rate Precision F-Measure

10-fold PS3 10-fold PS 10-fold PS 10-fold PS

Naive Bayes 67.86 64.76 .679 .648 .664 .647 .666 .642

Logistic 72.08 71.43 .724 .714 .712 .720 .713 .696

JRip 72.40 71.43 .714 .781 .711 .763 .711 .771

AdaBoostM1 71.43 70.48 .705 .781 .604 .763 .650 .771

LWL 72.08 70.48 .721 .705 .706 .683 .689 .677

1CCI: Correctly Classified Instances, 2TP: True Positive, 3PS: Percentage Split

Information
Retrieval

Information
Storage
.77**
Behavioral
Attitude
Intention
Information R2 = .33, e = .17 R2 = .59, e = .17
Sharing

Information
Application

*p< .01, **p< .001, Chi-Square= 393.25, DF= 235, Chi-Square/DF=1.67

Figure 1. The structural model

 ACCEPTED MANUSCRIPT

 Educational use of the MCC services in the higher education was investigated.
 A hybrid modeling approach combining both machine learning and SEM was employed.
 The structural model explained 59% of the variance in the behavioral intentions.
 Students’ information management practices were associated with the attitudes.
 OneR algorithm predicted the behavioral intentions with an accuracy of 74%.